Copyright (c) 1993 Texas Law Review Texas Law Review June, Tex. L. Rev LENGTH: words

Transcription

1 Copyright (c) 1993 Texas Law Review Texas Law Review June, Tex. L. Rev LENGTH: words SYMPOSIUM: REGULATING THE ELECTORAL PROCESS: Who's Afraid of Deliberative Democracy? On the Strategic/Deliberative Dichotomy in Recent Constitutional Jurisprudence. David M. Estlund * * Assistant Professor of Philosophy, Brown University. I have benefitted from discussions of this material with Alon Harel, Christopher Morris, Lewis Yelin, and the participants in the Texas Law Review Symposium, Regulating the Electoral Process, at The University of Texas School of Law, November 13-14, 1992, especially Frank Michelman. SUMMARY:... Deliberative democracy is an emerging ideal.... The ambiguous political valence of deliberative democracy is matched by the strategic conception's presence on both the left and the right.... This argument would, indeed, count in favor of deliberative democracy if the contrast were fair, but pluralism and the other main strategic conceptions of democracy are not so morally nihilistic.... If the category of deliberative democracy is to exclude such accomplished exponents of instrumental and strategic reason as Gauthier, then these stronger philosophical implications -- cognition, epistemology, and truth -- cannot be avoided.... Gauthier argues that "the outcome of deliberative politics is constitutive of justice among individuals" and "the common good." There is no independent standard of justice or common good to which constitutional deliberation aspires; the justice of the outcome is a fact solely about the procedure that produced it.... The mantle of deliberative democracy fits Rawls better than Gauthier for this reason.... An Epistemic Proceduralism of the kind I have sketched suggests that not only is this philosophically bolder deliberative theory of democracy truer to the most useful deliberative/strategic dichotomy, but eventually it may even be a viable theory of democratic legitimacy.... TEXT: [*1437] I. Introduction Deliberative democracy is an emerging ideal. The phrase was apparently coined only recently, n1 and it encompasses a cluster of views about politics that previously had long been unfashionable. The fashionable views, increasingly called "strategic" views of politics, characteristically applied doctrines or concepts developed primarily in welfare economics to the

2 actions of voters, political candidates, interest groups, and others in order to explain, predict, or morally evaluate behavior. n2 While much of that work has been done by economists and political scientists, the deliberative conception of politics has taken root especially in the fields of law and philosophy. In law, the deliberativists tend to have distinctive theoretical conflicts with practitioners (especially the early ones) of the thriving movement in law and economics. n3 In political philosophy, the deliberative theorists are [*1438] characteristically at odds not only with the orthodoxy in the social sciences, but also with those who conduct moral and political philosophy as an extension of the ideas of economic rationality and efficiency. n4 The deliberativists in political philosophy are also characteristically opposed to moral utilitarianism, n5 the influential view that morally ranks actions or policies according to their tendency to produce the greatest sum of individual well-being. n6 However, this may not go as deep as the other tendencies in the end: John Stuart Mill, a founder -- along with Jeremy Bentham -- of modern utilitarianism, n7 had a conception of democratic politics that would be categorized as deliberative by any sensible construal of the dichotomy. n8 In both legal and political theory, then, deliberative democracy is associated with a degree of opposition to the extension of classical economic methods and concepts to the study of politics. [*1439] There is no easy way to lump the deliberativists and the strategists into comfortable positions on the traditional left/right political continuum. Deliberative democracy's typical antipathy to laissez faire, or "invisible hand," n9 economic ideology in the study of politics tends to place it toward the left. n10 Its frequent embrace of some version of "civic virtue" n11 and community values, however, tends to place it toward the right. n12 The lesson is not that deliberative democracy is a centrist political philosophy. In certain versions -- especially certain leftist versions -- it has a clear noncentrist political valence. n13 However, nothing prevents more rightist versions from emerging in the mold of George Will's recent book. n14 The ambiguous political valence of deliberative democracy is matched by the strategic conception's presence on both the left and the right. Economics-oriented libertarians are not the only ones who conceive politics as a regulated rivalry among interest groups. Left-leaning work on racial polarization and voting rights reform often adopts the rhetoric, and perhaps also the normative framework, of strategic theories of democratic legitimacy. n15 It is in this context that the deliberative/strategic dichotomy has come to seem vulnerable. The dichotomy is most often drawn by theorists wishing to be placed on the deliberative side of the line. As a result, the deliberative side tends to get the better press, and even the better definition. In turn, theorists who fail to recognize their own views in the [*1440] unflattering accounts of strategic politics should be expected to don the mantle of deliberative politics, much as everyone has eventually adopted the well-regarded wardrobe of democracy itself over the last two centuries. A guiding question in this essay is whether the dichotomy between deliberative and strategic politics can be drawn so as to allow some self-respecting theorists to embrace each label. Why is this desirable? My own theoretical preferences would be well satisfied if everyone genuinely came to accept the ideal of deliberative democracy; there is no intrinsic value in having real partisans on each side. If this consensus is not forthcoming, though, different views of politics ought to be clearly distinguished, and I believe important differences can be captured

3 by a well-drawn dichotomy between deliberative and strategic politics. The dichotomy between strategic and deliberative politics ought to begin by defining strategic politics as a spare thing, driven solely or primarily by the employment of instrumental rationality. Individuals are supposed to begin with their diverse ends, desires, goals, or projects, and then to promote them as effectively as possible. n16 Strategy is a special case of instrumental rationality, namely the most effective promotion of one's ends in light of the fact that the results of one's choices will be partly determined by the choices made by other agents. Strategy does not imply competition; whether instrumentally rational agents will compete depends on whether one person's getting what she wants precludes other individuals from getting what they want. Where several people all prefer the same scarce goods, there will be competition. It is immediately clear, then, that strategic behavior will require individual deliberation in the employment of instrumental reason. Each individual will deliberate about which means to choose, and sometimes such deliberation will lead to coordinating actions with others to achieve greater benefit, and even to discussing with others how best to coordinate, or to persuading others that they have mistaken their interests by neglecting important information. Indeed, many actions may be aimed solely at the good of others, or even at the common good, depending on how highly these things rank in individuals' preferences. Individuals may behave in selfish, insensitive ways, or they may not. They might behave, instead, in cooperative, even apparently altruistic ways. It depends, for one thing, on how much they value such things as the well-being or autonomy of others for their own sake. An individual's behavior [*1441] will also depend on how beneficial it is to act selfishly or insensitively, as opposed to cooperatively or altruistically. If one's preferences are best served either directly or indirectly by it, a purely strategic agent might earnestly endeavor to support and comply with utterly egalitarian institutions and practices. This is all possible through the spare employment of instrumental reason in the face of other rational agents, that is, in the spare strategic conception of politics. How then to characterize deliberative politics? It certainly will not sufficiently define the concept to say that, unlike strategic politics, in deliberative politics there is deliberation or cooperation or the employment of reason, or all three. n17 Now it might be replied that while these things may be allowed as well by the idea of strategic politics, in practice the facts about scarcity and actual preferences show that purely strategic politics would indeed consist of unbridled selfishness, competition, and conflict. n18 But this claim is, of course, one of the more controversial propositions in the history of social thought. It is very much a live issue today. It should not be assumed without elaborate supporting argument that the distinction between strategic and deliberative politics is the distinction between conflict and cooperation, or between power and reason. These assertions are too close to saying it is the very distinction between evil and good. Maybe one of them leads, in actual fact, to more evil and less good than the other (a matter of legitimate controversy), but this should be an issue about theories that are first defined in other terms. It is better, then, to identify the strategic conception of politics with the insistence that the sole or predominant individual motives are those of instrumental rationality. This strategic conception can be contrasted, then, with all the other conceptions, those that give instrumental rationality something less than a predominant role. Among the many nonstrategic conceptions, how is the conception labelled deliberative democracy distinct?

4 Frank Michelman draws a distinction between two kinds of politics: deliberative and strategic. n19 David Gauthier puts Michelman's distinction [*1442] to heavy use in his recent theory of the authority of the Constitution. n20 It is important, then, to examine the strengths and weaknesses of Michelman's distinction and Gauthier's particular use of it. According to Michelman, deliberative politics is characterized by "argumentative interchange among persons who recognize each other as equal in authority" and as equally entitled to respect. n21 Each person may be persuaded by reasons relating to the claims of others as well as her own. A vote represents a "pooling of judgments" on some question of public ordering. n22 In contrast, strategic politics asks each participant to consider only her own interests. The outcomes of strategic politics, Michelman says, represent "not a collective judgment of reason but a vector sum in a field of forces." n23 Apparently, votes, like other political acts, are used simply to promote self-interest. n24 It is immediately clear which conception we should all favor. If these are the choices, what decent person could oppose the deliberative conception in favor of the strategic? The distinction is approximately between power and reason, between the view that might makes right and the view that politics is a sincere cooperative effort to promote the common good. Deliberative politics involves "argumentative interchange" aimed "in good faith" at "reasonable answers" while strategic politics yields not "reason" but "forces." n25 If I have accurately drawn the implications of a politics based on instrumental rationality, then we should expect any theorist to eschew strategic politics in Michelman's sense, since it leaves no room for reason, deliberation, or cooperation. Paradoxically, even a sophisticated instrumental-rationality theorist may feel more at home, with everyone else, in the deliberative camp, given Michelman's characterization of the choices. Cass Sunstein, another leading legal academic advocate of deliberative democracy, takes the main theoretical opponent to advocate "power" rather [*1443] than "reasons" in politics. n26 The "pluralist" view of politics n27 implies that might makes right. n28 This argument would, indeed, count in favor of deliberative democracy if the contrast were fair, but pluralism and the other main strategic conceptions of democracy are not so morally nihilistic. If nondeliberative politics is defined according to Sunstein as "[m]ight makes right" n29 or Michelman as the party of power (as opposed to the party of reason), n30 then all major political views will count as deliberative and their important differences will be missed. Sunstein uses one less tendentious formulation. In deliberative democracy, laws and policies must be justified by "public regarding reasons" and arguments. n31 Is it a public-regarding reason for a law to say that a majority of adequately informed citizens favor it? If so, it is difficult to think of a strategic theorist who would deny the importance of public-regarding reasons. If not, on what grounds? In other places, Sunstein associates deliberative democracy with a tendency to arrive at normative truths. "[A] large point of the [constitutional] system is to ensure discussion and debate... in a process through which reflection will encourage the emergence of general truths." n32 He approvingly cites James Madison's discussion of the system of national representation as a mechanism with which to "refine and enlarge the public views by passing them through the medium of a chosen body of citizens, whose wisdom may discern the true interest of their country." n33

5 In what follows, I hope to turn back David Gauthier's recent attempts to represent his own approach as deliberative, and to shore up the boundaries of this category in the process. To do so, it will be useful to adopt Michelman's placement of judgment at the core of deliberative politics n34 in a way that (so I will argue) puts the focus on cognitive, as distinct from [*1444] narrowly practical, reason. Deliberative politics has no monopoly on reason, but it does distinctively emphasize applying cognitive judgment, and therefore cognitive reason, to moral and political questions. Cognition, epistemology, and truth are all kindred concepts, and the fear they provoke in legal and political theory swells the ranks of the strategic theorists. n35 If the category of deliberative democracy is to exclude such accomplished exponents of instrumental and strategic reason as Gauthier, then these stronger philosophical implications -- cognition, epistemology, and truth -- cannot be avoided. I hope to show that much of the fear is misplaced, both philosophically and politically. There is a promising, though so far undeveloped, conception of democracy as epistemic. II. Gauthier David Gauthier theorizes about how a constitution comes to be the privileged embodiment of the popular will, and a law "higher" than ordinary legislation. n36 He argues that a constitution is a unanimous agreement about how to structure interaction so that strategic behavior will not lead to outcomes that are worse for all than other possible outcomes. n37 The problem is that of Hobbes: n38 pure strategic interaction where goods are limited will often lead to an outcome that is inferior in a compelling sense, namely that there is another possible condition that would be better for all (or at least better for some and worse for none). n39 The task is to devise a way for the participants to restructure their own interaction to remove this feature. Hobbes thought they could yield all their powers to a sovereign leader or state whose function it would be to make certain individual actions costly enough that each will be steered, by self-interest alone, to a collectively optimal outcome. n40 Locke famously pointed out one among many problems with this suggestion: A sovereign leader with unlimited powers is at least as great a threat to individual security as the original state of nature is, though neither is very secure. n41 If Hobbes's solution fails, [*1445] consider Gauthier's theory as a proposal to substitute a constitution for Hobbes's sovereign. How is it supposed to be an improvement? We know the ways in which the actual Constitution is superior to a Hobbesian sovereign, such as the separation of powers, the system of checks and balances, and the Bill of Rights, n42 but that should be put aside for now. We first need to know whether any constitution, and if so what kind, could be approved by Hobbesian individuals seeking to escape the state of nature. The theory is supposed to explain the authority of a constitution by reference to the way in which it was or would be chosen. Gauthier suggests that he is broadening the horizons of traditional Hobbesian subjects by making them "civic friends." n43 Hobbesian individuals regard each other as a threat, and for good reason. n44 Gauthier suggests that politics is impossible in such a case. n45 Suppose each person recognizes the potential for cooperation (as well as the potential threat) that other individuals represent. In this case there is the possibility of a genuinely political good. n46 It is unclear what is more "political" about this case, or why this is a good thing, as he clearly suggests. That aside,

6 the suggestion is that this interest in achieving cooperation is a key to the Hobbesian problem, a way out of mutually disadvantageous interaction. It is notoriously difficult for Hobbesian individuals to achieve cooperation because it is not clearly in anyone's interest to keep agreements to cooperate. Each person has an interest in getting others to "cooperate" but then taking the money and running. Each finds the other a convenient potential dupe. n47 It is hard to see how real cooperation would be possible among such agents. Gauthier has an elaborate theory, developed elsewhere, that it would be rational for such agents to, in effect, change their character to that of a noncheater. n48 Not only that, but Gauthier believes this is the correct account of the nature of morality: The dictates of morality are indirectly the dictates of rational self-interest. One would do well to do good, and to stop caring so much about doing well. The reason [*1446] is that cheaters are more or less detectable and cannot be trusted. They are left out of interactions from which they could have benefited even without cheating, and so they do worse than if they had not been (and been seen to be) cheaters. n49 Under these conditions, morality, being an implication of instrumental reason, is reconciled with reason. The theory is ingenious and, like any reasonably rigorous philosophical theory of morality, highly controversial. n50 In his discussion of the contractarian roots of the authority of a constitution, n51 Gauthier does not give any explanation of how cooperation is achieved. Instead, he skips over, without comment, the problem of how agreements to cooperate could be kept by self-interested agents. He simply assumes that they could, and suggests that in time such "bonds of convenience become ties of mutual civic concern." n52 Each "respect[s] the identity and aims of her fellows [and] willingly accord[s] them equal place in their common affairs with her own." n53 It is likely that the preceding account of how cooperation by self-interested agents is achieved borrows from the theory in Morals By Agreement. n54 There, Gauthier argues that a dimension of sociability can be "rationally reconstruct[ed]" from his account of the rational basis of morality. n55 A noninstrumental form of sociability exists in moral feelings themselves. Just because the account of morality is limited to considerations of instrumental rationality does not mean that moral feelings themselves must be narrowly instrumental: "In relating morality to the provision of benefits that themselves involve no affective concern with others, we do not thereby impoverish the moral feelings of persons who have such concern. It is because we can give morality a rational basis that we can secure its affective hold." n56 The constitutional deliberators apparently are individuals who are moral in the ways implied by Gauthier's moral theory. Accordingly, they subscribe to ideals of fairness and respect, and to this extent have a noninstrumental social regard for their fellows. n57 [*1447] On this reading, the constitutional theory utterly depends on the success of the moral theory. The traditional political problem about the cooperation of self-interested agents is presumed to be solved at a prior moral level of inquiry. That solution, ingenious and controversial, cannot be evaluated here. n58 But suppose it succeeds. We are still left with no clear description or derivation of the motivation of post-constitutional citizens, and no explanation of why a constitution would be a necessary constraint. Let us imagine, with Gauthier, that purely selfish individuals have transformed themselves into people who value every other comrade's interest as highly as their own. It is not clear what this means. It is clear

7 that for Gauthier such impartial motives are supposed to be present in the joint effort to devise a constitution. n59 By a "constitution" Gauthier means a way "to regulate their interaction to secure mutual benefit and express mutual respect." n60 But it is not clear what the problem is to which a constitution is a solution. If individuals have been transformed into utterly impartial agents, the Hobbesian problem of mutually disadvantageous interaction has been solved. Or, at least, there is no argument that it or anything like it would still obtain. n61 Unless Gauthier is assuming that the interactions of such individuals would still produce mutually worse outcomes, there is no problem for a constitution to solve. Gauthier is unclear about this issue: "I have supposed that those who first see others as useful will, as they share a wider range of activities, come to see those others as companionable, and will seek ways to regulate their interaction to secure mutual benefit and express mutual respect. These ways are embodied in their constitution." n62 The answer seems to be that impartiality does not come to characterize interaction in general, but primarily characterizes the choice of a constitution. However, the rest of interaction does not revert entirely to strategic, purely self-interested behavior, since it now willingly conforms itself to the dictates of the constitution. But within those constraints, interaction remains strategic or self-interested: "[I]n... constitutional politics, human [*1448] beings seek, through an appropriately conditioned strategic interaction, to direct and constrain their future interaction, which they anticipate to be largely strategic. They seek to remedy the failure of strategic interaction to yield outcomes affording fair mutual advantage." n63 It was obscure from the beginning how the transformation from selfishness to impartiality would come about. We simply imagined that cooperation was possible and then swallowed the psychological speculation that relations of convenience would, in time, produce a profound mutual respect leading to impartial concern of each for each. It now emerges that this (unspecified) psychological process leads individuals from relations of convenience to a willingness to regard all with utter impartiality in the choice of a constitution, but then largely to ignore any interests but one's own in all other interaction, except to confine one's actions to the dictates of constitutionally valid law. If morality is already being assumed, and it affords us the further assumption that each person regards everyone's interests impartially, then what warrants the conclusion that post-constitutional politics is largely strategic? Of course, post-constitutional politics is assumed to confine itself to the law as authorized by the constitution. Perhaps it is also assumed to confine itself to the dictates of morality, and then, within these combined constraints, to be strategically aimed at maximizing self-interest. It is unlikely that post-constitutional behavior is thought of as amoral apart from its fidelity to constitutionally authorized law; there would be nothing to warrant this assumption, and it would apparently be incompatible with the recommendations of the moral theory. n64 But with such robust constraints on maximizing behavior, it is not clear why further constitutional regulation would be required to avoid the inefficiencies of naked strategy. n65 Although Gauthier gives no reason to accept this complicated set of psychological premises, he hopes to ground the authority of the constitution on the fact that it would be agreed to by individuals who, out of enlightened self-interest, come to value everyone's interests equally. According to Gauthier, this is more than can be said for ordinary legislation, and that is what is "higher" about constitutional law.

8 III. Judgments and Independent Standards Gauthier, the leading philosophical exponent of the sufficiency of instrumental reason in moral and political theory, opts for deliberative [*1449] politics in Michelman's sense. Gauthier's conception of politics is not deliberative outside of the choice of a constitution. It is strategic, within the law as authorized by the constitution, in the rest of politics. Voting will conform to Michelman's description of voting in purely strategic politics. n66 Public political discussion, except when it concerns constitutional questions, will conform to the strategic model with the qualification that it will be conducted within the law. Ordinary politics is explicitly assumed not to be reliably deliberative: "If we were confident that these day-to-day decisions were themselves deliberatively reached, we should have no need of higher law." n67 Nevertheless, Gauthier claims that his account of constitutional politics is deliberative by Michelman's standards. He quotes and claims to comply with "Michelman's characterization: 'a reasoned interchange among persons who recognize each other as equal in authority and entitlement to respect, jointly directed towards answering some question of public ordering.'" n68 It is true that the characters that populate Gauthier's constitutional convention conduct themselves in distinctively unstrategic ways, if only because each gives equal weight to everyone's interests. Gauthier's constitutional designers are not constrained by justice or the common good. Rather, they give content to these ideas in their choice of a constitution. Gauthier argues that "the outcome of deliberative politics is constitutive of justice among individuals" and "the common good." n69 There is no independent standard of justice or common good to which constitutional deliberation aspires; the justice of the outcome is a fact solely about the procedure that produced it. n70 Is this sort of constitutive account compatible with the idea of deliberative politics? By insisting that justice and the common good are constituted by the constitutional deliberations, Gauthier makes it clear that there is no room for judgments about justice or the common good within the deliberations. This point does not depend on the claim, defended below, n71 that the idea of cognitive judgment involves objective standards. There is a logical breakdown of a sort if justice or the common good is held to be constituted by judgments about them. In general, the correct answer to a question cannot coherently be constituted by judgments as to the answer. n72 [*1450] Michelman says that in deliberative politics a vote represents a "pooling of judgments." n73 This assertion might be read as ruling out a politics that answers to no independent standard, a constitutivist view like Gauthier's. The use of the idea of judgments clears up an ambiguity involved in placing reason at the center of deliberative politics. Reasons can be reasons for choices, and some reasons for choices, those within "instrumental reason," are simple facts about the agent's ends and the means to achieving them. n74 Reason, in that sense, is not foreign to strategic politics. Many philosophers also believe there are other reasons for choices, and many locate moral reasons here. n75 Besides reasons for choices, there is another sense of reason, namely reasons to judge or believe that something is so. Philosophers sometimes call this the distinction between practical and theoretical reason, n76 although "theoretical" is not well chosen to reflect the contrast. We might speak of reason in the practical versus the cognitive sense. n77 It is not that there cannot be practical reasons to believe something; it is that practical reasons are not in general also cognitive reasons to believe. The same fact could be both a cognitive and a

9 practical reason. For example, if God tells me I will be damned if I lose faith, I have both a practical and a cognitive reason to believe. Practical for obvious reasons; cognitive, because if He tells me this, He must exist. n78 Michelman apparently means to characterize deliberative politics partly by the central importance it affords to cognitive reason, an importance not afforded in strategic politics. Unfortunately, the matter is complicated by a similar ambiguity in "judgment." We can form the judgment that capitalism is declining, and judgment in that sense will be true or false. But sometimes a judge makes [*1451] a judgment that someone shall be fined, or that someone shall have the legal status of guilty or not guilty. This type of judgment is not judgment in the cognitive sense, but in what I shall call the "performative" sense. Performative judgment does not admit of truth or falsity. Rather, to judge in the performative sense is to act so as to fix or determine the situation in certain respects. One might act this way based on beliefs or cognitive judgments that could be true or false. The act of performative judging, though, is not itself true or false. When Michelman speaks of pooling judgments he means judgment in the cognitive sense, not the performative sense. He means judgments that something is the case. Cognitive judgment on any topic involves the assumption that there is an objective standard. In judging that things are a certain way and not some other way, one presupposes that there is a way things are. Cognitive judgments may be inappropriate in certain contexts precisely because there are not facts of the matter (for example, whether Armstrong is better than Ellington), but that would not affect the point. If one holds that cognitive judgment is appropriate to deliberative politics, as Michelman appears to, then one thereby presupposes that there are objective standards of truth or correctness that apply to them. Objectivity is sometimes used in an imposing way, as entailing some or all of universality, necessity, timelessness, or independence. n79 By "objective" all we mean here is a standard by which individual judgments and the associated political outcomes could be mistaken. Michelman apparently holds that deliberative politics is distinctive in involving cognitive reason in a way strategic politics does not. If this reading of Michelman is correct, deliberative politics must involve the use of cognitive reason -- judgments that can be true or false. n80 The philosophical school known as pragmatism rejects many of the stronger metaphysical associations of the idea of truth. It has had some influence in legal thought, n81 and perhaps Michelman is thinking of it. Still, William James, a founding pragmatist, n82 even allows that "truth supposes a standard outside of the thinker." n83 Gauthier reads Michelman entirely differently. Michelman says that "[t]he deliberative attitude [does not] presuppose[] the existence of an [*1452] objectively best answer." n84 Gauthier apparently misinterprets this point. He notes that in Michelman's account, "[a] deliberative politics is characterized procedurally," as a certain kind of "reasoned interchange." n85 He concludes, fallaciously, that in general a deliberative politics in Michelman's sense is incompatible with the assumption of independent standards: "The appropriateness of the answers it yields... is established, not by any appeal to assumed expertise, but by the assurance that the manner in which it is conducted is informed by the standards that the answer must satisfy." n86 Michelman says nothing in the article cited by Gauthier to suggest this view. In a separate article, Michelman is clearer. He says, "It is true that notions of deliberative politics may be framed as

10 presupposing the existence of objectively discoverable, transcendently right or best answers," but they "need not" do so. n87 Michelman is in danger of making the inconsistent assertions that there need not be procedureindependent standards and that voting in deliberative politics is a "pooling of judgments." If judgments answer to no independent standard, then they are not judgments in the cognitive sense. Since performative judgments do not fit here, what kind of judgments are they? n88 Many people think democratic inputs are mere expressions of preference, not judgments at all. n89 They, of course, can consistently say that such inputs answer to no independent standard. Indeed the fact that the preference model enables one to avoid the need for independent standards is one of its main attractions for many theorists. n90 But Michelman's [*1453] conception of deliberative politics is designed to contrast with that view by concentrating on judgments instead of preferences, thus making it hard to see how independent standards can be avoided. I shall call a theory that allows justice or the common good to be constituted by political inputs constitutivist, and call all nonconstitutivist theories that nonetheless admit such things as justice or common good objectivist. n91 Among objectivist theories, some will see certain political inputs as cognitively addressing questions of justice or common good (for example, judging of them); I shall refer to these theories as cognitivist (and the remainder as noncognitivist). One could deny constitutivism without being a cognitivist. Cognitivism as employed here implies objectivism, but objectivism does not imply cognitivism. n92 Despite his remark that no objectively best answer is presupposed, Michelman's category of deliberative politics should not be read as simultaneously, and untenably, allowing constitutivist theories but allowing only cognitivist theories. Constitutivist theories cannot coherently be cognitivist. While Gauthier's theory cannot recognize independent standards of justice or common good prior to the choice of a constitution, nevertheless such a choice does answer to independent standards in a way. This lets his foot back in the deliberative door, but it also finally keeps him from getting through. IV. Is Gauthier's Theory Genuinely Deliberative? Under Gauthier's theory, parties to a constitutional convention deliberate and then unanimously agree to a set of institutions. In deliberating, they do not ask themselves what is just or what is to the common good, but what they choose will count as justice and the common good. They ask themselves what institutions will avoid the characteristic inefficiencies of unfettered strategic interaction. The deliberators are allowed to assume that ordinary (nonconstitutional) interactions will be largely strategic, but limited by whatever constitutional constraints are chosen. n93 It is fruitful to ask whether cognitive deliberation turns out to be entirely beside the point on this view. It is already clear that outside the choice (and amendment and interpretation) of a constitution, Gauthier's politics are largely strategic, though constrained by the law. The question [*1454] here is whether there is anything essentially deliberative about this theory of a constitution's authority. Suppose, to oversimplify for a moment, Gauthier's theory placed the authority of the constitution

11 in the hypothetical fact that if agents of a certain kind were situated in a certain way they would unanimously agree on a constitution of a certain kind. It is possible to show, first, that if his theory took this form, deliberation would be utterly irrelevant to the account in light of certain of its other features. The point will not just be that hypothetical deliberation does not count as deliberation; rather, such an account would not essentially involve even hypothetical deliberation. Second, while Gauthier's theory is explicitly not hypothetical in quite this way, its difference is not sufficient to incorporate deliberation in an essential way. I begin with the simplified, hypothetical interpretation of the authoritative constitutional choice. If the hypothetical fact about a certain kind of deliberation is what authorizes the actual constitution on Gauthier's view, then we should ask whether an equivalent justification could be given without any mention of deliberation or discussion. What I mean by an equivalent justification is sufficiently clear in the following, simpler example. The following two statements are equivalent in the pertinent respects even though one mentions agreement and deliberation and the other does not: Statement The best building is whichever building would be unanimously agreed, after deliberation, to be tallest by deliberators situated so as to have only accurate views about tallness, and having full information about all buildings. Statement The best building is whichever building is tallest. The mention of agreement and deliberation in a hypothetical agreement situation is not always removable in this way. Consider: Statement What is just is whatever would be agreed to after deliberation by individuals with different motives and preferences. Besides being "agreed to after deliberation," there is apparently no other salient property of the outcomes of such an agreement -- like tallness in the first two statements -- that would allow an equivalent statement of the principle. The kind of "removability" of agreement and deliberation in question, then, does not derive simply from the hypothetical form of an account. Reading Gauthier's condition as hypothetical for now, we can see that the constitutional "deliberators" would be addressing a question whose correct answer is independent of their deliberations, and they are assumed to have "full information." n94 The question they address is this: Which [*1455] institutional arrangement (if normatively accepted and coercively enforced) will avoid the inefficiencies that would be produced by unfettered strategic individual behavior? The answer is independent of their deliberations. They would consider the effects of each alternative set of institutions, and reject all those for which there is another institution that is better for all (or better for some and worse for none). They ask, that is, which arrangements are in the Pareto optimal set, and choose one of them (if there is more than one). Since the deliberators are assumed to have full information of the kind necessary to correctly answer the question about Pareto efficiency, the hypothetical deliberative formulation and a formulation that removes any mention of agreement or deliberation are equivalent:

12 Statement Justice is whatever set of institutions would be unanimously agreed, after deliberation by fully informed deliberators, to lead to more efficient outcomes than the results of unfettered strategic activity, or other available institutional arrangements. Statement Justice is whatever set of institutions would lead to more efficient outcomes than the results of unfettered strategic activity, or other available institutional arrangements. If the theory were entirely hypothetical in this way, then agreement and deliberation would be entirely removable from the account. Some readers will notice the similarity to a familiar point about the removability of agreement and deliberation in Rawls's "original position." n95 We will look at Rawls in this connection below. n96 A theory of democracy is not essentially deliberative unless it makes essential reference to processes of interpersonal deliberation. Gauthier's account of political justice violates this criterion since it would not be significantly changed if the hypothetical deliberative procedure were replaced by some nondeliberative procedure that was also a perfect efficiency detector. For example, consider a formulation that says justice is whatever would be printed out by a computer programmed to calculate the Pareto efficient options. Because the procedure infallibly returns the answer that is correct by a procedure-independent standard (a "perfect" procedure in Rawls's influential terminology), n97 any perfect procedure with regard to that standard will always return the same answer as Gauthier's hypothetical deliberative procedure. If the theory were [*1456] reformulated in any such way, it would be equivalent, in its results, to Gauthier's own formulation. Such theories would all call the same outcomes just or unjust; they would be coextensive in this respect. Suppose Gauthier replied that two theories that are coextensive, giving the same verdict in all cases, may yet be importantly different. After all, an arrangement's being efficient and its being correctly chosen as efficient are two different things. If this reply refers to an arrangement's actually being correctly chosen as efficient, it is surely a different thing from that arrangement's being efficient. But this idea is easily seen by noting that those two ideas are not coextensive. Some efficient arrangements will not be correctly chosen as efficient. However, if we choose truly coextensive properties -- (a) an arrangement's being efficient, and (b) an arrangement's being such that a hypothetical procedure that is perfect with regard to efficiency would choose it -- it is not clear that these are two different things. The two characterizations are certainly different in meaning in obvious respects -- as only one of them mentions agreement, for example. Still, it is not clear that they characterize different properties of arrangements because not only do they always agree in fact, but it is impossible for them to diverge under any possible circumstances. This is what I mean by saying they are not essentially different. Suppose Gauthier were to exploit the acknowledged difference in meaning between the two characterizations -- the one that mentions deliberation and the one that does not. The characterization that mentions a deliberative agreement among hypothetical individuals may not pick out a different property from characterizations that do not, but it brings out the moral significance of the single property captured by both characterizations. We can see how this property might plausibly be identified with the undoubtedly moral concept of justice by seeing how it can be characterized in terms of a certain kind of hypothetical rational agreement. Think

13 of it this way: It is very natural to see a close connection between justice and what would be voluntarily chosen by all the fully informed individuals who would be affected. Because we are dealing with a hypothetical case, there are no genuinely voluntary choices, and the closest we can come is to substitute rational and fully informed choices. n98 The claim that this hypothetical rational agreement is the closest we can come to voluntary choices requires some thought. Rawls proposes a hypothetical choice situation in which the parties are rational but, crucially, not fully informed because they are kept from knowing, and thus exploiting, [*1457] particular information about themselves or others. n99 Rawls argues that certain considerations are "arbitrary from a moral point of view," n100 and ought not to influence the outcome of a hypothetical choice designed to capture and clarify our convictions about justice. Gauthier never argues that this is not "getting closer" to the idea of justice as voluntary acceptance than his fully informed choice situation. Indeed, his primary methodological difference with Rawls is not that Rawls's choice situation is less morally plausible, but that it is less morally innocent -- it already needs to assume crucial moral convictions n101 in order to weave the "veil of ignorance," n102 whereas Gauthier hopes to derive moral convictions from the purely nonmoral apparatus of instrumentally rational choice. n103 It is almost certain that a theory without Gauthier's harsh constraints could get closer to the voluntariness conception of justice. For example, consider the view that what is just is what would be agreed to by fully informed and rational persons in light of all their knowledge, including their moral convictions (other than those specifically about justice). This is closer to the idea of a voluntary choice than either Gauthier's or Rawls's scheme, whatever its defects may be. By including a person's considered moral convictions, surely we would get closer to the idea of that person's voluntary choice, but Gauthier does not allow it. One needs antecedently to accept his more radical reductionist program before there is reason to accept the idea that rational fully informed agreement is as close as we can get to the idea of justice as voluntary agreement. To sharpen this point, we should ask in what respects the Gauthier agreement situation really brings out moral aspects of Pareto efficiency by seeing how it could be the object of a rational fully informed agreement in certain circumstances. Is this method in general a way of exposing an idea's moral aspects? If we can do this for Pareto efficiency, can we do it for other principles as well? We bring out the moral aspects of some principle p by showing how it could be deliberatively agreed to in a perfect p-detecting procedure. Does this work for the principle of, say, maximal entertainment value? Assume justice is whatever maximizes entertainment value. We imagine hypothetical individuals who are fully informed and motivated in the ways required to infallibly determine which social policies will maximize entertainment value. Next we note that certain policies would be deliberatively agreed to by individuals in this hypothetical choice situation. To do this analysis, of course, we would have to give the parties [*1458] different motives from those assumed by Gauthier. Then, compare the conclusions in the following statements, where the latter utilizes a formulation that leaves out deliberation and agreement: Statement Justice is whatever hypothetical individuals would deliberatively agree maximizes entertainment value. Statement Justice is whatever would produce the greatest sum of entertainment.

14 It is clear that in this example the deliberative formulation utterly fails to bring out any moral aspect of the principle that might have been suppressed in the nondeliberative formulation. This fanciful analogy may seem to ignore the possibility that the moral significance of Gauthier's formulation does not lie solely in the ideas of deliberation and agreement, but also in the assumptions about the parties' motives. In particular, at Gauthier's constitutional choice stage we are to assume (as we have seen) n104 that the participants are impartial as among their own utility and that of anyone else, as they are already moral individuals. Still, the fact that something would be agreed to even by moral individuals does not in general establish any moral significance of the thing chosen. Moral individuals can agree on morally indifferent as well as morally significant things. The question here is not whether Pareto efficiency is in fact a morally significant standard, n105 but whether, among a number of necessarily co-extensive characterizations of the standard, one that links it to the ideas of deliberation and agreement thereby captures any moral significance in the standard. If so, my charge that the theory is not essentially deliberative could be rebutted. As things stand, however, this reply appears not to succeed. Gauthier's theory is not essentially deliberative because its import is just as well captured by the nondeliberative standard of Pareto efficiency as it is by the idea of a Pareto efficiency-detecting deliberative agreement. This argument for excluding Gauthier's theory from the deliberative category depends on the assumption that the kind of deliberation that is central to that category is cognitive deliberation, as I have recommended. n106 If merely instrumental, noncognitive reason or deliberation were enough, then Gauthier's theory arguably would count as essentially [*1459] deliberative. The reason is that Gauthier's idea of an appropriate constitution is one that allows for ordinary interaction to be largely strategic without falling into characteristic Pareto inefficiencies. The account essentially involves strategic behavior, which intrinsically involves instrumental reasoning -- a noncognitive kind of deliberation. This type of deliberation is noncognitive in that its conclusions are not beliefs or propositions -- the kinds of things that can be true or false. Instrumental reasoning concludes in intentions or actions. If a theory like Gauthier's is not usefully categorized as deliberative, our emphasis on cognitive deliberation can help to explain why. So far, we have oversimplified by supposing Gauthier's standard for an appropriate constitution to be entirely hypothetical, that it would be chosen under certain circumstances. Gauthier's theory is, in fact, explicitly not hypothetical: "My account of the adoption of a constitution... does not treat it as merely hypothetical. The constitutional fact plays a key role in the argument." n107 However, the actual constitutional choice situation has its authority only "as a reasonable approximation" of the ideal constitutional choice situation. n108 Indeed, he says it is enough if some actual agreement may "be seen as making possible an interchange... the outcome of which might approximate to the outcome of a fair universal interchange." n109 It is not enough that an actual constitution accords with what (hypothetically) would have been agreed to in the ideal constitutional choice situation; it must also have been the actual product of some agreement that "represents itself as the expression of the people's will." n110 The requirement of an actual agreement is not sufficient, however, to bring deliberation back into